The service control layer of broadband metropolitan area network is a unique layer for service access network connecting with the core switched network, and the device which plays an important role is Broadband Remote Access Server (BRAS), which mainly undertakes the functions of service access control and user management.
As the assurance requirements for high quality services provided by the network have been put forward, higher requirements for the performance and reliability of the BRAS products in the service control layer are raised. The functions undertaken by the BRAS products and the locations of the BRAS products in the network determine that the BRAS has the networking with high reliability, therefore, the concept of dual-server hot backup of the BRAS emerges, which deploys two BRAS devices in the service control layer and makes it possess the functions of the main and standby redundancy and user information backup, so as to guarantee that a switchover and service recovery can be rapidly implemented during the device failure and achieve the service guarantee that they can't be sensed by the users. Meanwhile, the network hazards of the failure in the single point are also thoroughly solved.
The scenario of BRAS dual-server hot backup is deploying two BRAS products in the control layer of the network, and one of the two devices is in a main state and the other is in a standby state while working. The main device is used to control user access and record the user information, and backup the user information to the standby device through a way of real time synchronization or batch synchronization. When the main device is faulted, a main and standby switchover can be implemented rapidly, and since the user information has been stored in the device in standby state as backup, the user has no perceptions in the main and standby switchover process of the devices, which ensures the service quality and enhances the service experience of the user.
In the scenario of dual-server hot backup, there are two kinds of working modes: 1:1 and 1+1. It is assumed that the two BRAS devices are device A and device B, the working mode of 1:1 is targeted at all the users, and the device A is a device in main state and the device B is a device in standby state, that is, only the device A bears user services and the device B doesn't bear the users during the normal operation; the working mode of 1+1 is targeted at one set group of users, the device A is a device in main state and the device B is a device in standby state, and with respect to another set group of users, the device B is the device in main state and the device A is the device in standby state, that is, both the device A and device B bear the user services during the normal operation.
In the scenario of dual-server hot backup, two BRASs which are mutually backup devices have the same shared address pool. When the two BRASs work in the mode of 1+1, the users will gain access from two BRASs respectively and share the distributed addresses in the shared address pool simultaneously. When a certain user is online and obtains an address by distribution, the user information will be synchronized in real time to a remote backup device, and the remote device identifies the IP item in the local address record as unavailable according to the address distributed to the user in the synchronous information.
In the process of implementing the BRAS distributing an address for online user in the above 1+1 working mode, there is a following problem:
when there are users accessing two BRAS devices simultaneously, one BRAS distributes a certain address to a user, and before the BRAS synchronizes the address information of the user to a backup device, the backup device also has the users getting online and distributes the same address to a user, that is, two BRASs which are mutually backup devices distribute the same address for their respective user, which generates the address collision.